pymargins.LinearPredictionAdapter

class pymargins.LinearPredictionAdapter

Base for models whose prediction is exactly Xβ (no link function).

Covers OLS, WLS, GLS, IV, panel models. predict() is trivially Path A:

def predict(self, beta, X, offset=None):

return X @ beta + (offset if offset is not None else 0.0)

The complexity for these adapters lies in covariance(): handling the framework’s various vcov flavors (HC, cluster, HAC, GMM-style adjustments for IV, etc.).

__init__()

Methods

__init__()
attach(session)	Attach this adapter to a Margins session.
bootstrap_state()	Replay state for a refitted adapter.
coefficients()	Return β̂ as a 1D JAX array.
column_index_of_variable(name)	Return the design-matrix column index corresponding to a variable.
covariance([vcov_spec])	Return Σ̂ as a 2D JAX array.
design_matrix_from_df(df)	Build a design matrix from a concrete DataFrame of evaluation rows.
predict(beta, X[, offset])	Linear prediction: X @ beta + offset.
refit(resampled_data, *[, index])	Refit the model on resampled data, returning a new adapter.
variable_metadata()	Return per-variable metadata used by averaging and validation.

Attributes

gradient_backend_recommendation	Recommend pure autodiff for linear adapters.
n_outcomes	Number of outcome classes for multi-outcome models, default 1.
outcome_labels	Outcome class labels for multi-outcome models, or None.
supported_inference_methods	Linear adapters support delta, simulation, and bootstrap inference.
supports_jax_autodiff	Linear prediction is trivially JAX-differentiable (X @ beta).
training_data	The training data used to fit the model.